DE60132243T2 - METHOD FOR INDICATING ENTITIES - Google Patents

Description

The The present invention relates to a method of indexing for the Use with entities stored in a database.

It can easily be seen that if there is a huge number of entities in a database, identifying the entities in accordance with a query regarding the data in the database within a reasonable period of time is a non-trivial task. Around The recovery process is generally easier indexed the data in the database in some way, where the queries will then be executed at the index. It can be expected be that the way in which the entities are indexed, a significant Meaning of the quality and the speed of recovery possesses, and there information increasingly stored in databases, there is a significant interest to find improved ways to index data.

It is known to index location data based on place names. It is also known to retrieve a lot of geographic coordinates from the place names and to index based on the topological information extracted from the coordinates (eg "GIPSY": developed in UC Berkeley in the context of a common NSF / NASA / ARPA). Initiative (Wilensky et al., 1994)). In addition, it is known to install an index based on the geographic coordinates itself: the vendors of databases such. Oracle ^™ , have developed systems for storing and indexing geometric data - such as Oracle's spatial data cartridge, for example, which allows spatial querying to be performed using an extended (non-standard) form of SQL. Other vendors, such as MapInfo ^™ , SpatialWare ^™ , Innogistic ^™ and Informix ^™, have similar proprietary ways to handle spatial data. In particular, Innogistic ^{™ has developed} a product known as Cartology DSI which stores geometric vector data as blobs (large binary objects - which are not intrinsically recognizable by the underlying database). It also generates indexes outside the database based on the well-known quad-tree idea. The index data is stored in binary tree structures, accessed through middleware services of a distributed component object model (DCOM).

Both the Oracle ^™ and Innogistic ^™ systems use the quadrilateral tree process, which splits and subdivides a whole area of a layer into a series of four interlaced squares. The whole area is assigned to one of four squares, called 0, 1, 2, and 3. Each of these squares is divided into four smaller squares. The area of the square 1 becomes 10, 11, 12 and 13. Each of them is further subdivided, which means that e.g. For example, the subdivisions of the square 11 would be assigned the index values of 110, 111, 112 and 113. As a result, any location on the map can be referenced by a single index number. The disadvantage with this quadrilateral method is that processing time is wasted when there are no points in the divided squares; if the indexing is done over a large area, this wasted processing time is nontrivial and expensive.

US Patent US 6,161,105 describes a system by which multi-dimensional data is represented by binary hyperbubble (BH) code and which allows it to be indexed using the quadrilateral tree method.

One alternative indexing method is the k-dimensional binary search tree method by J.L. Bentley in "Multidimensional binary search trees for associative searching ", Commun. ACM 18 (9), 1975, 509-517 becomes. In this procedure, the vertical and horizontal Pages of a region containing points to index compared, taking the region through a straight line leading to the longer side is vertical, is divided into two parts. The position of the division or partitioning line is determined so that each resulting Part contains the same amount of data. This procedure is applied recursively to each of the two parts, until a predetermined criterion is met. Typically it stands this criterion with the capacity a page in a relational database (that is, the storage restrictions the database). The partitioned parts are called Node in a binary Tree, with the positions of the dividing lines between the parts the position of a node with respect to the other nodes in determine the tree. This procedure is for recovering points in terms of a section of interest particularly efficient because sections search space (binary Baums) by a simple comparison between the range of Interest and the dividing lines of the parts can be eliminated.

The Application of the k-dimensional method to geographic data becomes in "A file Organization for geographic information systems based on spatial proximity "of Matsuyama u. a, Computer Vision, Graphics and image processing, 26, 303-318, (1984), described.

Especially compare Matsuyama u. a. the efficiency of the k-dimensional tree method with the quadrilateral method, where they find that the k-dimensional Method both regarding the use of memory space (because the quadrilateral tree process typically quadrilaterals and thus memory Allocates regions where there are no points) as well in terms of data recovery (because the search space is efficient be reduced using the k-dimensional tree method can, as described above) a far more efficient method is.

regardless the fact that the k-dimensional tree method more efficient as the standard quad-tree method, it still rejects more areas to store than is actually required - as above wherein when a region is partitioned (Node in binary Tree), the partitioning line is determined that each resulting part contains the same amount of data; consequently the parts have a larger size than they would have to own finally the binary tree own more nodes than are necessary.

OOI, B. C. u. a. in Spatial KD-Tree: An Indexing Mechanism for Spatial Database, Proceedings of the Annual International Computer Software and Applications Conference (COMPSAC), Tokyo, 7-9. October 1987, Washington, IEEE Corp. Soc. Press., USA, Bd. Conf. 11, pages 433-438 the use of a so-called spatial kd-tree to processing the queries spatial Facilitate databases. The spatial kd tree is partitioned a set of records in k-dimensional space based on their spatial proximity in small groups.

According to one The first aspect of the present invention is a method for Attaching an index to multiple entities created in the Claim 1 of the attached Set of claims has defined features.

Consequently, the present "shrinks" Method compared to known indexing methods first the indexing space so that it only indexes over areas that Points included. Consequently, the memory areas are corresponding compact, which in terms of minimizing the use of memory space Has advantages.

In In fact, in the above-mentioned article entitled "A file organization for geographic information systems based on spatial proximity "Matsuyama et al. the problem associated with the quadrilateral tree process - namely Indexing empty pages of the database (see p. 312) - where but they solve the problem by applying a method similar to that of the present one Invention complete is different, (the k-dimensional tree method).

Advantageous For example, the step of shrinking the region includes calculating the distances between the entities in each of two dimensions; and for each of the dimensions that Identify which of the entities the greatest distance between them, so that when the two dimensions are vertical to each other, the region generated in step ii is a rectangle is defined by at least two points.

The Subregions can have one linked so-called "linked list" of points what is an effective linkage mechanism known in the art is.

Preferably The region is divided into four subregions, with the steps ii and iii for each of the second areas are recursively repeated.

Appropriately, the method contains saving the linkage data in a database, keeping a record of the number of entities which are stored in the database, and writing for each subregion the currently recorded number in their corresponding memory area.

• (a) Identifizieren einer oder mehrerer Regionen, die in dem vorgegebenen Bereich enthalten sind, wobei jede Region einen oder mehrerer Punkte umschließt und wobei jeder Region Verknüpfungsdaten zugeordnet sind, die für jede Region den Punkt oder die Punkte, die von dieser Region umschlossen sind, identifizieren; und
• (b) Zugreifen auf Verknüpfungsdaten, die den identifizierten Regionen entsprechen, um so Punkte, die von den identifizierten Regionen umschlossen werden, wiederzugewinnen.

Advantageously, the entities indexed in accordance with the first aspect of the invention can be recovered. In particular, and with respect to the points corresponding to the entities, the points contained in a given area can be recovered by performing the following steps:

• (a) identifying one or more regions included in the given area, wherein each region encloses one or more points and wherein each region is associated with linkage data identifying for each region the point or points enclosed by that region; and
• (b) accessing linkage data corresponding to the identified regions so as to retrieve points encompassed by the identified regions.

The Regions correspond to the memory areas that match are indexed with the first aspect of the invention. Consequently, because the storage areas are compact because, as described above they are "shrink wrapped" around the dots, the ID the points in the index an efficient process.

Preferably The identifying step (a) comprises the steps of retrieving a region and carrying out a process relating to the region, the process being the following Steps includes: Comparing the dimensions of the region with dimensions of the predetermined range to determine if the first region with overlaps the given area; and if there is an overlap exists, recovering subregions of the region and identifying any such subregions, the dimensions of which are completely within the predetermined range lie. This process can be for any subregion be repeated until all subregions thereof completely in fall within the given range are identified.

Be useful the multiple points in advance as a list of points in accordance with Relations between the respective regions and subregions thereof stored and contain the linkage data, a value the position of a first of the corresponding enclosed points in the list of points. The access step (b) comprises then for each of the identified regions and subregions: Recover an identifier representing the number of enclosed points, and Recovering a position value of the identified subregion assigned; and accessing the list of points and recoveries the number of enclosed points from a position in the list given by the position value.

Preferably corresponds to that in the linkage data included value of recorded recorded in the memory area Number.

The Process steps are preferably in one or more computer programs embodies that wholly or partly as a signal in a carrier wave for transmission over a communication network and / or in a computer readable medium as a computer program product personified could be.

Suitably correspond the dots data that are expressed in two dimensions can, z. B. Location data (length and latitude) or range data. The range data can run hours (eg opening and closing times), Price margins (eg maximum and minimum prices) and / or medical Data (eg, maximum and minimum blood pressure). Consequently, one could given range be a price range - such. B. a maximum house price and minimum house price. In accordance With the recovery method described above, the dimensions a given range (i.e., a price range) with one from the Index recovered region. All regions with overlap the price range, are then retrieved in turn until a region is identified is that completely falls into the specified price range. All the points in this thus fall within the identified region, thus representing goods which own a price that is in the specified maximum and minimum Price class falls.

Further Aspects, features and advantages of the present invention from the following description of the preferred embodiments of the invention, which refers to the attached drawing, wherein:

1 Figure 3 is a schematic diagram illustrating aspects of a communication system used by the invention;

2 is a schematic diagram showing an example of the points to be indexed according to the invention;

3 is a schematic diagram, which is an expanded view 2 shows;

4a u. 4b in combination, comprise a flowchart showing an embodiment of an indexing process according to the present invention, when the in 2 indexed points shown become;

5 is a schematic diagram showing the application of the process according to 4a u. 4b shows a square around the in 2 to produce points shown;

6 is a schematic diagram showing the application of the process according to 4a u. 4b shows to a sub-quad of the according to 5 generate generated quadrilaterals;

7 is a schematic diagram showing the application of the process according to 4a u. 4b points to a sub-quad of one of the sub-quads 6 to create;

8th an expanded view 7 is;

9 an expanded view 8th is that the application of the process after the 4a u. 4b shows another one in 7 to produce sub-quads shown;

10 an expanded view 8th is that the application of the process after the 4a u. 4b shows another one in 7 to produce sub-quads shown;

11 is a schematic diagram showing the application of the process according to 4a and 4b points to another sub-quad of one of the sub-quads 6 to create;

12 Fig. 12 is a schematic diagram illustrating a process of storing dots according to the invention;

13a u. 13b in combination, comprise a flowchart showing an embodiment of a retrieval process according to the present invention when the points are retrieved in accordance with a region of interest;

14 Fig. 12 is a schematic diagram showing an example of a region of interest for which points are to be recovered;

15 is a schematic diagram showing the application of the process according to 13a u. 13b on one of the sub-quads 6 shows;

16 and 17 an enlarged view 15 and are schematic diagrams illustrating the application of the process to the 13a u. 13b on a first of the sub-quads 7 demonstrate;

18 an enlarged view 15 is after and the area of interest and a second of the sub-quads 7 shows;

19 is a schematic diagram, the 18 corresponds and the application of the process after the 13a u. 13b on the second of the sub-quads 7 shows;

20a u. 20b schematic diagrams are showing the application of the process according to the 13a u. 13b to a (first) sub-quad of the in 18 show sub-quadrilateral shown;

21a u. 21b schematic diagrams are showing the application of the process according to the 13a u. 13b to another (a second) sub-quad of the in 18 show sub-quadrilateral shown;

22a u. 22b schematic diagrams are showing the application of the process according to the 13a u. 13b to another sub-quad (a third) of the in 18 show sub-quadrilateral shown;

23a u. 23b schematic diagrams are showing the application of the process according to the 13a u. 13b to another sub-quad (a fourth) of the in 18 show sub-quadrilateral shown;

24 an enlarged view 15 is and the area of interest and a third of the sub-quads after 7 shows;

25 is a schematic diagram, the 24 corresponds and the application of the process after the 13a u. 13b on the third of the sub-quads 7 shows;

26a u. 26b schematic diagrams are showing the application of the process according to the 13a u. 13b to a (first) sub-quad of the in 24 show sub-quadrilateral shown;

27a u. 27b schematic diagrams are showing the application of the process according to the 13a u. 13b to another (a second) sub-quad of the in 24 show sub-quadrilateral shown;

28a u. 28b schematic diagrams are showing the application of the process according to the 13a u. 13b to another sub-quad (a third) of the in 24 show sub-quadrilateral shown;

29a u. 29b schematic diagrams are showing the application of the process according to the 13a u. 13b to another sub-quad (a fourth) of the in 24 show sub-quadrilateral shown;

30 an enlarged view 15 is and the area of interest and a fourth of the sub-quads after 7 shows; and

31 is a schematic diagram, the 30 corresponds and the application of the process after the 13a u. 13b to the fourth of the sub-quads 7 shows;

32 is a flowchart showing further steps to follow in the recovery process 13a u. 13b are included; and each of the

33a - 33e Figure 3 is a schematic illustration of a region of interest for which points according to the invention are to be recovered.

overview

The database servers DB1, DB2, such as B. those in 1 typically store information for recovery by the users. At the physical level, the communication environment in which the database servers DB1, DB2 reside includes at least one user interface, which is generally provided by a computer terminal or workstation T3.

The embodiments of the invention executed in the workstation T3 which is connected to the database servers DB1, DB2. Even though the database servers DB1, DB2 are shown in the same LAN N1 as the terminal T3 are, of course, that the database server DB1, DB2 connected to other networks could be which in turn over one or more switches and / or routers (not shown) connected to the LAN N1. The embodiments receive data as input, z. As a file, and attach an index to the Data, as described in more detail below is described. The attached index will then be in one of the databases DB1, DB2 backed up, with the indexed data also in a given order by the structure of the attached index in one of the databases DB1, DB2. The attached one Index can be in the same or different database as the Database in which the data is stored.

The embodiments of the present invention relate to indexing entities defined by 2 dimensions. Obvious examples of the entities that may be indexed according to the embodiments include the location of objects, such as objects. As gas stations, ATMs, etc., since the position of the objects is generally defined in the form of width and length. Many other entities can be represented in 2 dimensions - eg. For example, the acceleration of a motorcycle as a function of time and speed, the conductivity of a material as a function of material properties and temperature, deformation of an article as a function of material properties, and force applied to the article, etc. In addition, transformations to n -dimensional parameters are used to reduce them to 2-dimensional parameters that can be shown in a 2-dimensional space.

One another example of entities, those using the embodiments of the present invention contains range information, z. As time information, pricing information and information about medical Conditions.

One Example of the time span information are the opening and closing times of business and recreational facilities - these Times can expressed in two dimensions be, for. B. with the closing and opening times on the ordinate or abscissa axes. Similarly, delivery times (earliest and latest) expressed in two dimensions become.

One Example of price information contains the prices of goods, so that z. B. maximum and minimum prices of goods each in one expressed in two dimensions and in this way using the embodiments of the invention can be indexed. The pricing information also includes trade information, how they are used to transfer securities, stocks, bonds etc. to buy and sell.

One Example of information about medical conditions contains Statistics regarding measurable states, such as B. Body temperature and blood pressure, and states that are translated into numerical representations can, such as For example, the sites of carcinomas.

In In the following description, the entities are generally referred to as "points" to the context of the entities (eg, gas station, ATM, etc.) of the mechanics of the embodiment to separate.

An overview of an embodiment of a method for indexing points will be given with reference to FIG 2 described. 2 shows a region R1 in which several points 200 are located. Each of the points is defined in a space of x, y coordinates. The region R1 comprising points to be indexed is substantially examined and divided into the regions R3, R4 containing points and the region R2 containing no points (the regions could be divided into any shape such as Rectangle, triangle or strip, to be parted; 2 Figure 4 shows the areas that are divided into strips to describe the inventive concept of this invention). The embodiment then examines the distribution of the points in the regions R3 and R4, and then, on a scale smaller than that considered for the region R1, identifies the regions in R3 and R4 that comprise points. In 3 R4 becomes substantially R11, and the distribution of points in R11 is examined. By concentrating the distribution of points in this way, the areas not containing points become the region R2 in FIG 2 and the R12 region in 3 , implicitly discarded. The process is continuously repeated, effectively "down" through a succession of areas of decreasing size until the size of a region is such that it collapses to the size of a single point. As the embodiment "deepens down," each area is linked to the area above it so that each point is linked by a series of areas. An index for these points includes the sequence of ranges, which ranges and points are used to create an index (to be described in more detail below) that is saved in database DB1. The relationship between the points and the regions allows the points to be identified by identifying an area in the index.

One of the advantages of creating an index, as described above, is that the query search area is limited to areas known to contain points - ie, the queries are executed only on the areas saved in database DB1 , and since these areas by definition contain points, the search area is relatively compact. Back in 2 For example, the process of identifying points with respect to a query according to the above-described embodiment is faster than if the index includes information on the whole area R1.

In a particular form of an embodiment, which is described below with reference to the 4 - 11 is represented, the points are 2-dimensional coordinates in x, y space. If the entities to be indexed are acceleration values defined by a corresponding set of time and velocity values, the time and velocity values are mapped directly to the space of the x, y coordinates such that (t1, v1), (t2, v2), ..., (tn, vn) are coordinates of points corresponding to the acceleration values. If the entities to be indexed are place values defined by a corresponding set of latitude and longitude values, the latitude and longitude values become direct are mapped onto a space of x, y coordinates such that (lat_1, long_1), ..., (lat_n, long_n) are the coordinates of points corresponding to the location values. It is assumed that the points have been stored (eg written to a file) so that the embodiments of the invention read the points in the form of a file. In alternative embodiments, a user may enter the points when the index is applied to these points.

In addition, in the following embodiment, the areas are squares, which will be referred to as "square" and "sub-quad" in the following description, each square being divided into four sub-quadrants in order. Each sub-quad is examined for the presence of points. Those without dots are discarded, which is an equivalent process for discarding the area R2 described above with reference to FIGS 2 and 3 has been described, with each new sub-quad, containing points, being "shrink-wrapped" around the smallest area containing dots in that sub-quad. (The embodiment analyzes the areas in accordance with squares, but many other shapes could be used to perform the "shrink wrap" around the points.) Each sub-quad is then subdivided into four sub-quads, with empty sub-quads discarded again and every one remaining Sub-quad around its smallest area containing dots is "shrink wrapped". Finally, any remaining sub-quad is "shrink-wrapped" to a single point, with its coordinates being that of the point of interest. Once the quad and all sub-quads, including both the intervening sub-quads that have not been discarded and the sub-quads that coincide with individual points, have been identified, an index for the points that includes the quad and sub-quads will be identified are relevant for everyone. This is after the discussion of 5 - 10 described in detail.

Hereinafter, steps S4.1 to S4.9 are sequentially included in the flowchart of FIGS 4a and 4b shown and through the operations with the same reference numerals as used in the 5 - 11 are shown illustrated.

5

• • Step S4.1 Read in the x, y coordinates of all points to be indexed (As stated above, the points of any Storage location, such. A file, or read directly by a user become);
• • Step S4.2 Create a bounding box for all points, identify and perform the "shrink wrapping" around the coordinates of the outermost points (the bounding box is the difference between the coordinates of the outermost points in both the x and y dimensions given: dx or dy). The bounding box is the outermost square 501 ;
• • Step S4.3 u. Step 4.4 Secure the dimensions of the quadrilateral 501 - ie dx, dy - and the coordinates of the points in it;
• • Step S4.5 Check if the outermost square 501 has a positive size (ie, dy, dy of the quadrilateral 501 equals zero?). In the in 5 example shown there is an extreme quadrangle 501 because there are several points in the quadrilateral, dx, dy are different from zero, hence the quadrilateral 501 has a positive size;
• • Step S4.6 Divide the outermost square 501 in four sub-quads 503a . 503b . 503c . 503d ;
• • Step S4.6.1 Label each point with its relevant sub-quad and record the number of dots in each sub-quad;
• • Step S4.7 Check starting with the sub-quad 0 ( 503a ), if there are any points in the sub-quad 503a gives.

If there are points, give the points in this sub-quad 0 ( 503a ) in step S4.1 and run for the sub-quad 0 ( 503a ) continue through the steps step S4.1, as described below with reference to 6 is described.

6

• • Step S4.1 Read in the x, y coordinates of the points that correspond to the sub-quad 503a correspond;
• • Step S4.2 Create a bounding box for all points in 503a , create a "shrink-wrapped" sub-quad 0 601 , This illustrates the principle described above - the embodiment identifies an area in sub-quad 0 in which there are no points, this area is then discarded;
• • Step S4.3 u. Step S4.4 Save the dimensions of sub-quad 0 ( 601 ) - ie dy, dy - and the coordinates of the points in it;
• • Step S4.5 Check if the sub-quad 601 has a positive size? (That is, dx, dy of the quadrilateral 601 equal to zero?) As in 6 can be seen, are in the square 601 several points are dx, dy for the quadrilateral 601 different from zero and thus has the sub-quad 601 a positive size;
• • Step S4.6 Divide the sub-quad 0 601 in 4 sub-quads; 0.0 ( 603a ), 0.1 ( 603b ), 0.2 ( 603c ), 0.3 ( 603d );
• • Step S4.6.1 Label each point with its relevant sub-quad and record the number of dots in each sub-quad;
• • Step S4.7 Check starting with the sub-quad 0,0 ( 603a ), whether in the sub-quadrant 0.0 ( 603a ) there are any points: Since there are points, these points in this sub-quad give 0.0 ( 603a ) in step S4.1 and run for the sub-quad 0,0 ( 603a ) continue through the steps step S4.1, as described below with reference to 7 is described.

7

• • Step S4.1 Read in the x, y coordinates of the points that correspond to the sub-quad 603a correspond;
• • Step S4.2 Create a bounding box for all points in the sub-quad 0,0 ( 603a ), create a "shrink-wrapped" sub-quad 0,0 701 , As for the sub-quad 0, the area without dots in the sub-quad becomes 603a ignored;
• • Step S4.3 u. Step S4.4 Save the dimensions of the sub-quad 0,0 ( 701 ) - ie dy, dy - and the coordinates of the points in it;
• • Step S4.5 Check if the sub-quad 701 has a positive size? (That is, dx, dy of the quadrilateral 701 equal to zero?) As in 7 can be seen, are in the square 701 several points are dx, dy for the quadrilateral 701 non-zero and has this sub-quad 701 a positive size;
• • Step S4.6 Divide the sub quad 0,0 701 in 4 sub-quads; 0,0,0 ( 703a ), 0,0,1 ( 703b ), 0,0,2 ( 703c ), 0,0,3 ( 703d );
• • Step S4.6.1 Label each point with its relevant sub-quad and record the number of dots in each sub-quad;
• • Step S4.7 Check starting with the sub-quad 0,0,0 ( 703a ), whether it is in the sub-quad 0,0,0 ( 703a ) there are any points: there is a point in the sub-quad 0,0,0 ( 703a Therefore, enter these points in this sub-quad 0,0,0 in step S4.1 and run for the sub-quad 0,0 ( 703a ) continue through the steps step S4.1, as described below with further reference to 7 is described.

Also in 7

• • Step S4.1 Read in the x, y coordinates of the points that correspond to the sub-quad 703a correspond;
• • Step S4.2 Create a bounding box for all points in the sub-quad 0,0,0 ( 703a ), create a "shrink-wrapped" sub-quad 0,0,0: here around a single point;
• • Step S4.3 u. Step S4.4 Secure the dimensions of the "shrink-wrapped" sub-quad 0,0,0, which now points to one single point down so that dy, dy = 0, and sure the coordinates of the point;
• • Step S4.5 Check if the sub-quad has a positive size? (That is, dx, dy of the point equals zero?) In fact, dx and dy are both zero because the sub-quad 703a collapsed into a single point. Therefore, to step S4.8;
• Step S4.8 Increment the sub-quad counter i at this level (0,0, i), give the points (step S4.7) in the sub-quad 0,0,1 ( 703b ) in step S4.1 and run for the lower quadrant 0,0,1 ( 703b ) continue through the steps step S4.1, as described below with reference to 8th is described.

8th

• • Step S4.1 Read in the x, y coordinates of the points that correspond to the sub-quad 703b correspond;
• • Step S4.2 Create a bounding box for all points in the sub-quad 0,0,1 ( 703b ), create a "shrink-wrapped" sub-quad 0,0,1 801 , As for the sub-quad 0, the area without dots in the sub-quad becomes 703b ignored;
• • Step S4.3 u. Step S4.4 Save the dimensions of the sub-quad 801 - ie dy, dy - and the coordinates of the points in it;
• • Step S4.5 Check if the sub-quad 801 has a positive size? (That is, dx, dy of the quadrilateral 703a zero?) in the sub-quad 801 if there are several points, dx, dy are for the quadrilateral 801 different from zero and thus has this sub-quad 801 a positive size;
• • Step S4.6 Divide the sub-quad 0,0,1 801 in 4 sub-quads; 0,0,1,0 ( 803a ), 0,0,1,1 ( 803b ), 0,0,1,2 ( 803c ), 0,0,1,3 ( 803d );
• • Step S4.6.1 Tag each point with its relevant sub-quad and draw the number of Points in each sub-quad on;
• • Step S4.7 Check starting with the sub-quad 0,0,1,0 ( 803a ), whether it is in the sub-quadrant 0,0,1,0 ( 803a ) There are no points in the sub-quadrant 0,0,1,0 ( 803a );
• Step S4.8 Increment the sub-quad counter i at this level (0,0,1, i) and give the points (step S4.7) in the sub-quad 0,0,1,1 ( 803b ) enter step S4.1 and run for the sub-square 0,0,1,1 ( 803b ) continue through the steps step S4.1, as described below with reference to 9 is described.

9

• • Step S4.1 Read in the x, y coordinates of the points that correspond to the sub-quad 803b correspond;
• • Step S4.2 Create a bounding box for all points in the sub-quad 0,0,1,1 ( 803b ), create a "shrink-wrapped" sub-quad 0,0,1,1, which is a single point;
• • Step S4.3 u. Step S4.4 Ensure the dimensions of the single point - d. H. dy, dy - and the coordinates of the point;
• • Step S4.5 Check if the point has a positive size? (That is, dx, dy of the point equals zero?) In fact, dx and dy are both zero because the sub-quad 803b collapsed into a single point. Therefore to step 4.8;
• Step S4.8 Increment the sub-quad counter i at this level (0,0,1, i) and give the points in the sub-quad 0,0,1,2 ( 803b ) enter step S4.1 and run for the sub-quad 0,0,2 ( 803b ) continue through the steps step S4.1, as described below with further reference to 9 is described.

Also in 9

• • Step S4.1 Read in the x, y coordinates of the points that correspond to the sub-quad 803c correspond;
• • Step S4.2 Create a bounding box for all points in the sub-quad 0,0,1,2 ( 803c ), create a "shrink-wrapped" sub-quad 0,0,1,2, which is a single point;
• • Step S4.3 u. Step S4.4 Ensure the dimensions of the single point - d. H. dy, dy - and the coordinates of the point;
• • Step S4.5 Check if the point has a positive size? (That is, dx, dy of the point equals zero?) In fact, dx and dy are both zero because the sub-quad 803c collapsed into a single point. Therefore, to step S4.8;
• • Step S4.8 Increment the sub-quad counter i at this level (0,0,1, i). There are no points (step S4.7) in the sub-quad 0,0,1,3 ( 803b Therefore, back to step S4.8: increment the sub-quad counter i at this level (0,0,1, i): but i> 3, therefore
• Step S4.9 Return to the sub-quad level 0,0, i and increment the sub-quad counter from 1 to 2 and thus consider the sub-quad 0,0,2 ( 703c ): There is one point in the sub-quad 0,0,2 ( 703c ), therefore the points (step S4.7) in the sub-quadrant give 0,0,2 ( 703c ) in step S4.1 and run for the sub-quad 0,0,2 ( 703c ) continue through the steps step S4.1, as described below with reference to 10 is described.

10

• • Step S4.1 Read in the x, y coordinates of the points that correspond to the sub-quad 703c correspond;
• • Step S4.2 Create a bounding box for all points in the sub-quad 0,0,2 ( 703c ), create a "shrink-wrapped" sub-quad 0,0,2, which is a single point;
• • Step S4.3 u. Step S4.4 Ensure the dimensions of the single point - d. H. dy, dy - and the coordinates of the point;
• • Step S4.5 Check if the point has a positive size? (That is, dx, dy of the point equals zero?) In fact, dx and dy are both zero because the sub-quad 703c collapsed into a single point. Therefore to step 4.8;
• • Step S4.8 Increment the sub-quad counter i at this level (0,0, i). There are no points (step S4.7) in the sub-quad 0,0,3 ( 703d Therefore, back to step S4.8: increment the sub-quad counter i at this level (0,0, i): but i> 3, therefore
• Step S4.9 Return to the sub-quad level 0, i and increment the sub-quad counter i from 0 to 1 and thus consider the sub-quad 0,1 ( 603b ), as with reference to 11 is described.

11

The in the 8th - 10 The process described is repeated, but for the sub-quadrant 0.1, where once all points within the sub-quad O, n sub-quads have been assigned, the process goes to sub-quad 1.

As described earlier, attaching an index at these points is then a matter of backing up the sub-quad information. This can be done in many ways, but preferably the index includes the sub-quad information saved in steps S4.3 and S4.4, namely the dimensions of the sub-quad (in the x, y coordinates) and the coordinates of those falling into it Points and a link to the 4 subquadrants in the sub-quad. Thus, the index essentially comprises a hierarchy of sub-quad structures, the hierarchy being given by the relationship between each successive sub-quad and its 4 sub-quads. The sub-quads are ordered in accordance with the sub-quad hierarchy of the largest sub-quad (here 501 in 5 ) written down to the smallest sub-quad in the index.

In addition to securing the sub-quad structures in database DB1, the points are placed in the database (either in the same database or another database) in an order given by the inverse of the sub-quad hierarchy, e.g. In a file. Thus, in this case, the dots in the sub-quads at the bottom of the hierarchy are first written to the file. Since the points are written to a file, a running calculation of the total number of points (recording of the number of points) is maintained so that a counter representing the current number of points that has been encountered so far, + 1, into the corresponding one Subarray structure is written as each point is written to the file. The calculation operates from the smallest sub-quadrilateral, essentially indicating, for each sub-quadrilateral, the position of the first of all points in that sub-quad, compared to all indexed points (Pos _{sub_quad} ), e.g. B. with reference to 12 , sub-quad Points in the sub-quad (N) Number of the first point written in the sub-quad (Pos _{sub_quad} ) highlighted point in the point file 0,0,1,1 n (N = 1) 1 n, m, p, o, ... 0,0,1,2 m (N = 1) 2 n, m, p, o, ... 0,0,0 p (N = 1) 3 n, m, p, o, ... 0,0,1 n, m (N = 2) 1 n, m, p, o, ... 0,0,2 o (N = 1) 4 n, m, p, o, ... 0.0 n, m, p, o (N = 4) 1 n, m, p, o, ... 0.1, c, d some points 5 n, m, p, o, next ...

Thus, the dot file for the sub-quad is 501 (the outermost quadrangle, see 5 ) n, m, p, o, ... (starting from the smallest sub-quad 0,0,1,1). Since both the number of points (N) and the number in the current calculation of the points (Pos _{sub_quad} ) corresponding to the first point in a sub-quad are written in the sub-quad, then once a sub-quad of interest has been identified , the points lying in the identified sub-quad are extracted by going to (Pos _{sub_quad} ) in the point file and extracting N points from that position. This is demonstrated in an embodiment demonstrating the recovery of points.

Recovering information

The The second invention relates to a method for recovering of entities by means of an index of elements related to the entities stands if the relationship between each element and the other Elements in the index and the relationship between the elements in the index and the indexed entities are known. The process is easy on one in accordance index generated by the method of indexing outlined above applicable because the index is a hierarchy of sub-quadrilateral structures includes and the hierarchy (through the quadrilateral> quadrilateral relations) well defined. However, it should be remembered that that Method is equally applicable to any type of index, that meets these conditions.

• • die Elemente im Index mehrere Bereiche in einem vorgegebenen Bereich sind,
• • die indexierten Entitäten Punkte sind, die durch zwei Dimensionen definiert sind;
• • es eine vorgegebene Beziehung zwischen den Bereichen gibt; und
• • es eine vorgegebene Beziehung zwischen den Bereichen und den Punkten gibt.

In the following description, an embodiment of the retrieval process will be described in which an index is queried for the points with a query specifying a "region of interest". It is assumed that:

• • the elements in the index are several areas in a given area,
• • the indexed entities are points defined by two dimensions;
• • there is a given relationship between the areas; and
• • There is a given relationship between the areas and the points.

• a) mit der Region von Interesse überlappt und
• b) in diesen Bereichen Punkte enthält, die mit der Region von Interesse überlappen.

In essence, the embodiment identifies which of the areas

• a) overlapped with the region of interest and
• b) contains points in these areas that overlap with the region of interest.

The given relation between the areas and the points then used to extract the points that were identified in the Areas fall.

A region of interest refers to a region in the entity's two-dimensional representation. Thus, for the time span described above (the closing and opening times of business and leisure facilities), a region of interest would be a period of interest - such B. "the shops open between 10.00 and 13.00 clock". The region of interest will then be defined by a region (preferably a square) bounded within pre-specified points. In the 33a -E the region of interest could be any of the following:
(0, 10), (13, 24) = the facilities, which sometime between 10.00 u. 13.00 are open ( 33a )
(0, 13), (10, 24) = the facilities that run continuously between 10.00 u. 13.00 are open ( 33b )
(0, 10), (10, 13) = the facilities that are open before 10.00 but close before 13.00 ( 33c )
(10, 13), (13, 24) = the facilities that open after 10.00 but close after 13.00 ( 33d )
(10, 10), (13, 13) = the facilities that open after 10.00 u. close before 13.00 ( 33e )

Similarly, for the above described price range being a region of interest a price range so for a retrieval request of "all the elements that are somewhere in the Range of £ 50.00 and £ 80.00 fall, "the region of interest could be defined by a region defined in the points (0, 50) and (80, 80) is limited.

For the location information described above, a region of interest would be an area of locations, such as location. B. "all elements that are between a first position ((lat, long) ₁ ) and a second position ((lat, long) ₂ )".

A flowchart showing the steps of a method for identifying areas that overlap with the region of interest when the region of interest is related to location information is disclosed in U.S. Patent Nos. 4,774,866 13a and 13b shown. The steps are then for the in 14 shown region of interest in the 14 - 31 illustrated. In this embodiment, specific examples of the regions in a given region are referred to as quadrangles and sub-quads. In the 13 The method steps shown below are described below with reference to each of 14 - 31 described.

14

• • S13.1.1 Read the x, y coordinates of a region of interest (x1, y1), (x2, y2). As an example could this, if a user wishes the motor vehicle workshops to locate in a specific area, as latitude / longitude pairs be indexed defining points in a two-dimensional space where the geographical area in which the user is interested is as a "region of interest "in can be expressed in two-dimensional space;
• • S13.1.2 Win the coordinates of the points and the size of the square for the outermost square again and set (X1_Q, Y1_Q), (X2_Q, Y2_Q) to the size of the outermost one quadrangle;
• • S13.2 Assess if the region of interest requires pruning: if x1 <X1_Q, set x1 = X1_Q; if y1 <Y1_Q, set y1 = Y1_Q; if x2> X2_Q, set x2 = X2_Q; if y2> Y2_Q, set y2 = Y2_Q. For the in 14 Example region of interest shown does not satisfy any of these conditions;
• • S13.3 Judge if x1> x2 or y1> y2. In this case, no conditions are met;

• • S13.4 Beurteile, ob die Region von Interesse exakt mit dem Viereck überlappt. Nein:
• • S13.5 Gewinne ein Unterviereck (503a) des aktuellen Vierecks (501) in Übereinstimmung mit S13.1.2 wieder, wie unter Bezugnahme auf 15 beschrieben ist.

Steps S13.2 and S13.3 are just examples of conditions that can be used to determine if the sub-quad recovered in S13.1.2 overlaps with the region of interest, and if there is any overlap, any points in the overlapping region; it is contemplated that alternative methods could be used to determine this.

• • S13.4 Judge if the region of interest overlaps exactly with the quadrangle. No:
• • S13.5 Win a sub-quad ( 503a ) of the current quadrilateral ( 501 ) in accordance with S13.1.2 again, as with reference to 15 is described.

15

• • S13.1.2 Get the coordinates of the points and the size of the sub-quad from the "shrink-wrapped" sub-quad 0 601 Again, set (X1_Q, Y1_Q), (X2_Q, Y2_Q) to the size of the "shrink-wrapped" sub-quad 601 ;
• • S13.2 Assess whether the region of interest requires pruning: if x1 <X1_Q, set x1 = X1_Q; if y1 <Y1_Q, set y1 = Y1_Q; if x2> X2_Q, set x2 = X2_Q; if y2> Y2_Q, set y2 = Y2_Q. In this case, none of these conditions are met;
• • S13.3 Judge if x1> x2 or y1> y2. In this case, no conditions are met;
• • S13.4 Judge if the region of interest overlaps exactly with the sub-quad. No:
• • S13.5 Recover a sub-quad of the current sub-quad in accordance with S13.1.2 as described with reference to 16 and 17 is described.

16 u. 17

• • S13.1.2 Get the coordinates of the points and the size of the sub-quad from the "shrink-wrapped" sub-quad 0,0 701 Again, set (X1_Q, Y1_Q), (X2_Q, Y2_Q) to the size of the "shrink-wrapped" sub-quad 701 ;
• • S13.2 Assess whether the region of interest requires pruning: x2> X2_Q, so set x2 = X2_Q, and y2> Y2_Q, therefore set y2 = Y2_Q;
• • S13.3 Judge if x1> x2 or y1> y2. In this case ( 17 ), both conditions are met, meaning that there are no points in the sub-quad 0,0 ( 701 ) that fall into the region of interest;
• • S13.6 Increment the sub-quad counter i at this level (0, i) and recover the sub-quad (0,1) in accordance with S13.1.2 as described with reference to FIG 18 and 19 is described.

18 u. 19

• • S13.1.2 Take the coordinates of the points and the size of the sub-quad from the "shrink-wrapped" sub-quad 0,1 again: set (X1_Q, Y1_Q), (X2_Q, Y2_Q) to the size of the "shrink-wrapped" sub-quad 0,1;
• • S13.2 Assess whether the region of interest requires pruning: x1 <X1_Q, therefore set x1 = X1_Q, and y2> Y2_Q, therefore set y2 = Y2_Q;
• • S13.3 Judge if x1> x2 or y1> y2. In this case, no conditions are met;
• • S13.4 Assess whether the (now cropped) region of interest is exact overlaps with the sub-quad. No:
• • S13.5 Recover a sub-quad 0,1,0 of the current sub-quad 0,1 in accordance with S13.1.2, as with reference to 20a and 20b is described.

20a u. 20b

• • S13.1.2 Take the coordinates of the points and the size of the sub-quad from the "shrink-wrapped" sub-quad 0,1,0 again: set (X1_Q, Y1_Q), (X2_Q, Y2_Q) to the size of the "shrink wrapped" subarray 0,1,0;
• • S13.2 Assess whether the region of interest requires pruning: x2> X2_Q, so set x2 = X2_Q, and y2> Y2_Q, therefore set y2 = Y2_Q;
• • S13.3 Judge if x1> x2 or y1> y2. In this case ( 20b ) both conditions are met, which means that there are no points in the sub-quadrant 0,1,0 that fall into the region of interest;
• • S13.6 Increment the sub-quad counter i at this level (0,1, i) and (S13.6.1) recover the sub-quad (0,1,1) in accordance with S13.1.2 as described with reference to FIG 21a and 21b is described.

21a u. 21b

• • S13.1.2 Take the coordinates of the points and the size of the sub-quad from the "shrink-wrapped" sub-quad 0,1,1 Again, set (X1_Q, Y1_Q), (X2_Q, Y2_Q) to the size of the "shrink wrapped" subarray 0,1,1 - d. H. a single point;
• • S13.2 Assess whether the region of interest requires pruning: x1 <X1_Q, therefore set x1 = X1_Q, and y2> Y2_Q, therefore set y2 = Y2_Q;
• • S13.3 Judge if x1> x2 or y1> y2. In this case ( 21b ) both conditions are met, meaning that there are no points in the subarray 0,1,1 that fall into the region of interest;
• • S13.6 Increment the sub-quad counter i at this level (0,1, i) and (S13.6.1) recover the sub-quad (0,1,2) in accordance with S13.1.2 as described with reference to FIG 22a and 22b is described.

22a u. 22b

• • S13.1.2 Recover the coordinates of the points and the size of the sub-quad from the "shrink-wrapped" sub-quad 0,1,2: set (X1_Q, Y1_Q), (X2_Q, Y2_Q) to the size of the "shrink wrapped" subarray 0,1,2 - d. H. a single point;
• • S13.2 Assess whether the region of interest requires pruning: x1 <X1_Q, therefore set x1 = X1_Q, y1 <Y1_Q, x2> X2_Q and y2> Y2_Q, so set x1 = X1_Q, y1 = Y1_Q, x2 = X2_Q, y2 = Y2_Q;
• • S13.3 Judge if x1> x2 or y1> y2. No;
• • S13.4 Assess whether the (now cropped) region of interest is exact overlaps with the sub-quad: Yes;
• • S13.4.2 Draw the sub-quad and the number of dots in the sub-quad (here 1);
• • S13.6 Increment the sub-quad counter i at this level (0,1, i) and (S13.6.1) recover the sub-quad (0,1,3) in accordance with S13.1.2 as described with reference to FIG 23a and 23b is described.

23a u. 23b

• • S13.1.2 Take the coordinates of the points and the size of the sub-quad from the "shrink wrapped" subarray 0,1,3 Again, set (X1_Q, Y1_Q), (X2_Q, Y2_Q) to the size of the "shrink wrapped" subarray 0,1,3 - d. H. a single point;
• • S13.2 Assess whether the region of interest requires pruning: x1 <X1_Q, therefore set x1 = X1_Q, y1 <Y1_Q, x2> X2_Q and y2> Y2_Q, so set x1 = X1_Q, y1 = Y1_Q, x2 = X2_Q, y2 = Y2_Q;
• • S13.3 Judge if x1> x2 or y1> y2. No;
• • S13.4 Assess whether the (now cropped) region of interest is exact overlaps with the sub-quad: Yes;
• • S13.4.2 Draw the sub-quad and the number of dots in the sub-quad (here 1);
• • S13.6 Increment the sub-quad counter i at this level (0,1, i) ... i> 3, therefore (S13.6.2) recover the sub-quad (0,2) in accordance with S13.1.2, as with reference to the 24 and 25 is described.

24 u. 25

• • S13.1.2 Take the coordinates of the points and the size of the sub-quad from the "shrink-wrapped" sub-quad 0,2 again: set (X1_Q, Y1_Q), (X2_Q, Y2_Q) to the size of the "shrink-wrapped" sub-quad 0,2;
• • S13.2 Assess whether the region of interest requires pruning: y1 <Y1_Q, therefore set y1 = Y1_Q, and x2> X2_Q, therefore set x2 = X2_Q;
• • S13.3 Judge if x1> x2 or y1> y2. In this case, no conditions are met;
• • S13.4 Assess whether the (now cropped) region of interest is exact overlaps with the sub-quad. No:
• • S13.5 Recover a sub-quad 0,2,0 of the current sub-quad 0,2 in accordance with S13.1.2 as described with reference to 26a and 26b is described.

26a u. 26b

• • S13.1.2 Recover the coordinates of the points and the size of the sub-quad from the "shrink-wrapped" sub-quad 0,2,0: set (X1_Q, Y1_Q), (X2_Q, Y2_Q) to the size of the "shrink-wrapped" sub-quad 0,2,0;
• • S13.2 Assess whether the region of interest requires pruning: x2> X2_Q, so set x2 = X2_Q, and y2> Y2_Q, therefore set y2 = Y2_Q;
• • S13.3 Judge if x1> x2 or y1> y2. In this case ( 26b ) both conditions are fulfilled, which means that there are no points in the sub-quadrant 0.2.0 that fall into the region of interest;
• • S13.6 Increment the sub-quad counter i at this level (0,2, i) and (S13.6.1) recover the sub-quad (0,2,1) in accordance with S13.1.2 as described with reference to FIG 27a and 27b is described.

27a u. 27b

• • S13.1.2 Take the coordinates of the points and the size of the sub-quad from the "shrink-wrapped" sub-quad 0,2,1 Again, set (X1_Q, Y1_Q), (X2_Q, Y2_Q) to the size of the "shrink wrapped" subarray 0,2,1 - d. H. a single point;
• • S13.2 Assess whether the region of interest requires pruning: x1 <X1_Q, therefore set x1 = X1_Q, y1 <Y1_Q, x2> X2_Q and y2> Y2_Q, so set x1 = X1_Q, y1 = Y1_Q, x2 = X2_Q, y2 = Y2_Q;
• • S13.3 Judge if x1> x2 or y1> y2. No;
• • S13.4 Assess whether the (now cropped) region of interest is exact overlaps with the sub-quad: Yes;
• • S13.4.2 Draw the sub-quad and the number of dots in the sub-quad (here 1);
• • S13.6 Increment the sub-quad counter i at this level (0,1, i) ... i <3, therefore (S13.6.1) recover the sub-quad (0,2,2) in accordance with S13.1.2 as with reference to the 28a and 28b is described.

28a u. 28b

• • S13.1.2 Take the coordinates of the points and the size of the sub-quad from the "shrink-wrapped" sub-quad 0,2,2 again: set (X1_Q, Y1_Q), (X2_Q, Y2_Q) to the size of the "shrink-wrapped" sub-quad 0,2,2;
• • S13.2 Assess whether the region of interest requires pruning: x2> X2_Q, so set x2 = X2_Q, and y1 <Y1_Q, therefore set y1 = Y1_Q;
• • S13.3 Judge if x1> x2 or y1> y2. In this case ( 28b ) both conditions are met, which means that there are no points in the subarray 0,2,2 that fall into the region of interest;
• • S13.6 Increment the sub-quad counter i at this level (0,2, i) and (S13.6.1) recover the sub-quad (0,2,3) in accordance with S13.1.2 as described with reference to 29 is described.

29

• • S13.1.2 Take the coordinates of the points and the size of the sub-quad from the "shrink-wrapped" sub-quad 0,2,3 again: there is no sub-quad equivalent to 0,2,3 because it is there are no points in the range corresponding to this sub-quad, therefore jump to S13.6;
• • S13.6.2 Increment the sub-quad counter i at this level (0,2, i): i> 3, therefore (S13.6.2) retrieve the sub-quad (0,3) in accordance with S13.1.2, as referenced on the 30 and 31 is described.

30 u. 31

• • S13.2 Assess whether the region of interest requires pruning: x1 <X1_Q, therefore set x1 = X1_Q, y1 <Y1_Q, x2> X2_Q and y2> Y2_Q, so set x1 = X1_Q, y1 = Y1_Q, x2 = X2_Q, y2 = Y2_Q;
• • S13.3 Judge if x1> x2 or y1> y2. No;
• • S13.4 Assess whether the (now cropped) region of interest is exact overlaps with the sub-quad: Yes;
• • S13.4.2 Draw the sub-quad and the number of dots in the sub-quad (here 2);
• • P13.6 Increment the sub-quad counter i at this level (0, i) ... i> 3, therefore, (S13.6.2) win the sub-quad 1 in accordance with S13.1.2 again.

How out 14 is apparent, the region of interest falls exclusively in the sub-quadrilateral 0 and the sub-quadrilateral therein, so that when the in the 13a and 13b is applied to the sub-quads 1, 2 and 3, the conditions applied in steps S13.2 and S13.3 cause the process to end within a few steps.

At the end of the process of identifying the sub-quads that overlap with the region of interest, the sub-quads and the number of points in those sub-quads recorded in steps S13.4.2 are returned - for this example:
Sub-quadrant 0,1,2, number of points: 1 point;
Sub-quadrant 0,1,3, number of points: 1 point;
Lower quadrant 0,2,1, Number of points: 1 point;
Sub-square 0.3, number of points: 2 points.

Once the sub-quads have been identified, the actual points are retrieved. In this embodiment, and as described above, the dots are in a flat file stores. In addition, each sub-quad structure stores a number representing the position of the first point in a respective sub-quad relative to the total number of indexed points (see, eg, FIG. 4 , all points in the square 501 ) indicates.

• • S32.1 Gewinne für dieses Unterviereck die Anzahl der Punkte wieder, die in das Unterviereck fallen, (N);
• • S32.2 Gewinne die Position des ersten Punkts von der entsprechenden Unterviereck-Struktur (Pos_{sub_quad}) wieder;
• • S32.3 Bewege einen Dateizeiger in eine durch Pos_{sub_quad} gegebene Position in der Punktedatei;
• • S32.4 Extrahiere aus dieser Position N Punkte aus der Datei.

The process for actually recovering the points is in 32 shown:
For each quadrangle recorded in step S13.4.2:

• • S32.1 For this sub-quad, recover the number of points that fall into the sub-quad, (N);
• • S32.2 Recover the position of the first point from the corresponding sub-quad structure (Pos _{sub_quad} );
• • S32.3 Move a file _pointer to a position in the point _file given by Pos sub_quad;
• • S32.4 Extract N points from the file from this position.

implementation

The in the 4a and 4b , the 13a and 13b and 32 The processes described are implemented in software and are executed in one of the terminals T3, T4 or are distributed between the terminals T3, T4. The terminals T3, T4 thus represent one of several computers, preferably server computers.

The to be indexed points can be over a File or similar in the terminals T3, T4, the index that has been generated, such as described above can be stored in the database DB1 and the point file can also be stored in database DB1 get saved. An area of interest may be in the form of a Database query entered in a known manner via a (not shown) client terminal entered and over the network N1 to the terminals T3, T4 transferred becomes.

Preferably are the processes described above in the C programming language implements and uses recursive programming techniques, to "lower down" in the sub-quads to subdivide. It goes without saying that such a method for the invention is immaterial.

Further details and modifications

As set forth above, the invention can be used to index and retrieve data in 2 dimensions expressed are. The invention can also used to index data in more than 2 dimensions and regain, provided the data (n-dimensional Data, where n> 2 applies) be transformed in 2 dimensions. In such cases, the transformed 2-dimensional data according to the invention indexed and be recovered. Objects in a 3-dimensional space are defined z. B. using a package such. B. NCAR Graphics, what a unix-based Graphics package is that a wide range of capabilities for the Display and manipulation of numerical data offers and of the University Corporation for Atmospheric Research is to be transformed in 2 dimensions. (See http://www.dkrz.de/ngdoc/ng4.0.1 for information in terms of NCAR Graphics and http://ngwww.ucar.edu/ngdoc/ng/fund/chp16-21/threed.html for information in terms of the aspects of 3-to-2-dimensional transformation.)

It could other variations executed become. It could z. B. simply dividing the squares and sub-squares into a different number of areas at each iteration, such as B. eight or ten instead of four, can be used.

Claims (15)

A method of attaching an index to a plurality of entities, wherein each entity is represented by data identifying a point defined in a space by generating a hierarchy of subregion data structures, the method comprising: i. Creating a minimum constraint region around the points representing the entities (S4.2); ii. Performing a process related to the generated region or subregion, the process comprising: generating a region or subregion data structure; Storing data specifying the extent of the respective region or subregion in the region or subregion data structure; Storing links to points that fall within the respective region or subregion, in the area or subrange data structure (S4.3); Dividing the region or subregion to create multiple subregions (S4.6); and for each of the plurality of sub-regions containing points: reducing the sub-region until its further reduction would cause one of the points to fall outside the sub-region; and iii. repeatedly executing the process with respect to each generated subregion containing points until a predetermined criterion is satisfied (S4.5); the method further comprising storing links to sub-area data structures corresponding to their respective plurality of subregions in each area or sub-area data structure; and wherein the hierarchy of the subarea data structures provides the index to the entities. The method of claim 1, wherein the predetermined The criterion is that a subregion contains a single point. Method according to one of the preceding claims, wherein the step of shrinking the region includes: To calculate of distances between points, the entities correspond, in each of two dimensions; and for each of Dimensions Identify which of the points is the greatest distance between them, so that if the two dimensions to each other are vertical, the generated region is a rectangle that points through is defined, which correspond to at least two entities. Method according to one of the preceding claims, wherein the region is divided into four subregions. Method according to one of the preceding claims, wherein the steps ii. and iii. recursively repeated. Method according to one of the preceding claims, which includes: Saving the entities in a database, hold a record of the number of entities stored in the database are for each subregion and writing the number currently recorded into its corresponding subarea data structure. The method of claim 6, wherein entities in the Database in accordance with the size of a subregion be written so that entities that are in the smallest subregions fall, first be written to the database. Method according to one of the preceding claims, which receiving a trigger signal to call the steps i. to iii. includes. Method according to one of the preceding claims, which and recovering one or more entities that are included in a predetermined range, wherein the or any entity indexed by the method of any one of the preceding claims which comprises the following steps: Identify one or more regions included in the given area are, each region enclosing one or more points and wherein each region linkage data are assigned for each region the point or points that enclosed this region are, identify; and Accessing linkage data, corresponding to the identified regions, so points that be surrounded by the identified regions to regain. The method of claim 9, wherein the identifying step comprises the steps of: retrieving data identifying a region; Performing a process related to the region, the process comprising the steps of: comparing the extents of the region with extents of the predetermined region to determine if the region overlaps the predetermined region; and if there is an overlap, retrieving data identifying subregions of the region and identifying any such subregions whose extents are entirely within the predetermined range; and for each subregion, repeating the process until all subregions thereof that fall entirely within the predetermined range are identified, wherein the accessing step comprises accessing linkage data corresponding to the identified subregions so as to regain points enclosed by the subregions. The method of claim 10, wherein the plurality of points in advance as a list of points in accordance with relationships stored between respective regions and subregions thereof and the linkage data contain a value representing the position of a first of the corresponding indicates enclosed points in the list of points, and where the access step for Each of the identified regions includes: Regaining one Identifier representing the number of enclosed points and retrieving a position value of the one identified Region is assigned; and Access the list of points and retrieving the number of enclosed points from one Position in the list given by the position value. Method according to one of claims 9 to 11, wherein the regions and the corresponding linkage data stored in an index. Method according to one of the preceding claims 9 to 12, wherein the predetermined range corresponds to range information. The method of claim 13, wherein the range information geographical range information, operating hours information, Delivery time information and / or medical information include. Computer program product containing a set of commands contains to cause a computer to follow the steps of the procedure according to the claims 1 to 14.